I was using yadif to deinterlace. I understand that a smart deinterlacer combines two fields to produce one full frame, which results in reducing the fps in half, but at least full vertical resolution is maintained.
But now, I was searching for a higher quality deinterlacer, and I found QTGMC, but I'm very confused about how this deinterlacer maintains the full vertical resolution. If source is 25i and has 576px vertical resolution, and QTGMC output is 50fps, then all frames will have 288px vertical resolution upscaled and interpolated to 576px ? Or does it copy fields from other frames, for the current frame, in order to maintain 576px vertical resolution ?
Because Yadif combines two half frames to make one full frame, and I want to make sure that, if I switch to QTGMC, vertical resolution will not be reduced. What deinterlacer will drop more from the vertical resolution, yadif or qtgmc ?
+ Reply to Thread
Results 1 to 4 of 4
Thread: qtgmc and vertical resolution
If you're asking which of these smart bobbers usually produces higher quality output, it's QTGMC. How can you know for sure? Use it. Caution: be certain your source is interlaced, not telecined.Our inventions are wont to be pretty toys, which distract our attention from serious things. They are but improved means to an unimproved end. -- Henry David Thoreau
mode = 0 : temporal and spatial interlacing check (default). = 1 : double framerate (bob), temporal and spatial interlacing check.Use it.
Consider a still shot. Any two consecutive fields are complementary and can simply be woven together to make a complete frame with no loss of resolution.
In a shot with a static background but something moving the foreground -- the background can be woven together as in a still shot. Only the moving parts of the image in the foreground need to be addressed further. With simple motions, like the object moved four pixels to the left, the scan lines can be filled by shifting the other field right by four pixels. Only with very complex motions (where QTGMC can't figure out how to use another field to fill the missing scan lines) will the filter have to resort to a technique that requires interpolation from only one field, losing resolution.